Predicting Temperature and Humidity for Cotton Field Using Deep Learning Models in Smart Agriculture System

Agriculture demonstrates a pivotal part in sustaining world economies as it proves to be the backbone of human civilization. Cotton crops, contributing to the textile industry and the livelihoods of millions of farmers worldwide, ensure food security and hold immense importance for a country's economy. There are many imposed challenges for the farmer due to unpredictable weather conditions, making it a problem in making informed decisions about planting, crop management and harvesting. Extreme temperature and weather conditions can lead to loss of crop and yield causing financial instability for farmers. To counter these challenges, an artificial intelligence (AI) enabled Smart Agricultural System based on deep learning (DL) models is deployed to assist farmers in providing optimal cotton crop management. The main objectives of conducting this research involve overcoming the challenges of poor crop management through effective usage of DL models to make accurate predictions regarding weather conditions for crop maintenance. The DL models are trained using data such as temperature and humidity. Different models such as convolutional neural networks (CNN) model, long short-term memory (LSTM) model and the most powerful model convolutional neural network - long short-term memory (CNN-LSTM) are implemented, and a fair comparison is done through evaluating the most suitable and accurate performing model in terms of prediction. The insightful evaluation metrics errors are root mean squared error (RMSE), mean squared error (MSE) and the mean absolute error (MAE). The results of this study demonstrate promising outcomes as per the used evaluation metrics. The adaptation of powerful deep learning models in an AI-enabled system provides optimal crop management for farmers through accurate prediction of weather parameters, leading to better crop quality and yield. The integration of machine learning and deep learning techniques enable data-driven approaches in agriculture, showcasing the value of technological advancements and sets the stage for future innovations in this sector.